Valproate toxicity in a child: two novel observations

To the Editor:

Sodium valproate, at therapeutic levels (350–700 micromol/L; 50–100 mg/L), is largely protein bound (80%–94%). However in overdose the protein binding is saturated and higher proportions of the drug are free. The elimination of half-life increases in overdose from 6 to 16 hours to over 30 hours. In neonates and infants the elimination of the drug is also low, although this increases in children to levels higher than in adults per kilogram body weight.^1,2

The cardiac effects of valproate overdose described are tachyarrhythmias and heart block.³ No effects on the corrected QT intervals have been reported to our knowledge.

A 2-year-old girl, weighing 12 kg, presented to her local emergency department following an accidental overdose of sodium valproate (non-modified release) estimated to be 4 g (330 mg/kg). Although initially she had a slight reduction in conscious level with a Glasgow Comas Scale of 13/15, this subsequently dropped to 8/15, necessitating intubation for airway protection, and admission to our paediatric intensive care unit.

Haemodialysis has been shown to be effective in the clearance of valproate in overdose.⁴ However haemodialysis is not readily available on our, or most U.K., paediatric intensive care units, out of hours. Our patient's initial valproate level was measured at 4277 micromol/L (611 mg/L) 8 hours post-ingestion. It was also noted that her QTc on ECG was 500 ms, confirmed manually using Bazzett's formula. She was normothermic at the time and on a low dose of noradrenaline for blood pressure support. We established her on continuous veno-venous haemofiltration with pre-dilution, a flow rate of 100 ml/min and a filtration fraction of 15%–30%. After 19 hours of filtration the valproate level was measured at 679 micromol/L (97 mg/L), i.e. within the therapeutic range. Haemofiltration was discontinued. Twelve hours after stopping haemofiltration the valproate level was 506 micormol/L (72 mg/L) suggesting continuing clearance. Her QTc had now reduced to 400 ms. Her Glasgow Coma Score was 14/15 with some continuing confusion. However she was recognising both parents and communicating at discharge to her local hospital for continuing care.

Using the three time points the elimination constant (K_e) while on CVVH was 0.097 per hour, giving a half life (t_1/2) of 7.16 hours; off CVVH the K_e was 0.016 per hour and t_1/2 43.32 hours. In comparison, the K_e quoted by Aly⁴ using haemodialysis and haemoperfusion in series, in an adult case, with comparable valproate levels was 0.19 per hour, giving a t_1/2 of 3.6 hours.

We accept that we have data from only three time points and no valproate levels from the ultra-filtrate fluid. Nevertheless, given the clinical improvement in our patient and an accelerated elimination of plasma valproate, haemofiltration has a place in the management of valproate overdose in children, especially if haemodialysis is not readily available. Also in our patient the QTc, noted to be high with elevated valproate levels, returned to normal (< 450 ms) once the valproate levels became therapeutic. Although monitoring is not recommended routinely for children on valproate, the possibility of prolonged QTc and its consequences may be relevant in valproate overdose.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.